Behavior modification research, theory, and practice hold great promise for the development of efficient and economical learning-based procedures for the treatment of stuttering. The author's detailed review of work to date indicates the following: (1) Stuttering appears to decrease in frequency when any event whatever is made contingent upon it, suggesting in conjunction with other evidence that operant learning manipulation should be addressed to some more basic response in the stutterer's repertoire than non-fluency; (2) By considering jaw muscle tension as a basically important maladaptive response in stutterng, a variety of stuttering-related research findings can be nicely integrated; (3) The role of anxiety as a basic mediating aspect in maintaining stuttering is questionable, although the precise way in which anxiety contributes to the problem is not clear at this time; (4) There are case reports of varying degrees of success in the behavioral treatment of stuttering, but no adequately controlled research studies exist. The research proposed is addressed to the first three points. (1) The effect of contingent and random positive and aversive events on nonfluency will be examined within the context of a single set of experimental procedures, providng a direct comparison of the results of various manipulations. (2) Electromyographic potentials from stutterers' jaw muscles while reading and speaking will be continually displayed to the subject in a biofeedback design, and reinforcement will be made contingent on tension reduction, in an attempt to teach stutterers to reduce tension; and the effect of this learning on stuttering severity will be examined. This is the major part of the proposed work, which will also proceed as far as possible on the development of an automated biofeedback procedure for stuttering therapy.